VM0047 is the most rigorous baseline in carbon. It still cannot price the risk.

In carbon, the methodology is usually where the argument ends. A project picks a standard, the standard sets the rules for counting tonnes, and the credits that come out the far end carry the rules' authority. So when Verra published VM0047 for afforestation, reforestation and revegetation in October 2023, it was reasonable to read it as the closing of a long-running integrity gap. It was not. It was the best methodology the sector has produced for planting carbon, and it still cannot price the risk that capital actually holds.

What VM0047 actually fixed

01 · The end of the developer's baseline

For two decades the central weakness of afforestation and reforestation crediting was the baseline. Under the old Clean Development Mechanism methodologies that VM0047 replaces, the counterfactual was modelled. A developer projected what the land would have done without the project, and the project was credited against that projection. The projection was an assumption, and the party making the assumption was the party paid by the result. VM0047's first and most important move is to take that assumption away and replace it with an observation.

In the area-based approach, the project area is matched to a set of control plots: comparable unmanaged land, selected to resemble the project on historical trajectory, and then tracked through time using remote sensing. The benchmark is not a forecast of what the land might have done. It is a continuously updated reading of what comparable land is actually doing. The crediting baseline becomes the change in a stocking index on those controls, and the project is credited only on the gap between its own measured stocking and theirs. Verra calls this a dynamic performance benchmark, and it is genuinely the right idea.

02 · How the count is built

The stocking index is a remote-sensing proxy for vegetation, typically a spectral measure correlated with above-ground biomass and then calibrated against field plots on the ground so that an index value maps to a tonnage. The benchmark is re-tested at every verification, not fixed once at the start, which means additionality and the crediting baseline are re-examined against live evidence over the project's life rather than asserted at the outset. For smaller and dispersed planting, VM0047 offers a census-based approach instead, which counts trees directly and, under strict eligibility, sets the baseline conservatively at zero. The methodology also credits fractionally rather than in a single block, so a project earns in proportion to the change it can actually demonstrate.

The deepest move is what it nets out. Older approaches quietly credited natural regeneration, growth that would have happened anyway, as if the project had caused it. By benchmarking against controls that are themselves regenerating, VM0047 removes non-project regeneration from the count. That single design choice is the difference between selling additional carbon and selling the weather.

Why netting out the counterfactual is not a detail

03 · The Saloum Delta problem

The case for dynamic benchmarks is not theoretical. In the Saloum Delta in Senegal, Andrieu and colleagues combined botanical field study with remote sensing and concluded that the mangrove system had largely recovered on its own, driven by the return of rainfall. Their 2020 paper in Forest Ecology and Management found that roughly 96 per cent of mangrove regeneration was spontaneous and linked to rainfall recovery, with planting accounting for the small remainder. Yet large mangrove restoration crediting in the same landscape was built on the premise that the trees were there because they had been put there.

A modelled baseline lets you sell the rainfall. A dynamic benchmark makes you prove the planting.

That is the precise failure VM0047 is engineered against. If the controls are regenerating because the rain came back, the project is credited only on what it adds beyond that recovery. The methodology turns a structural over-crediting risk into a measured quantity. This is real progress, and it deserves to be recognised as such rather than folded into a general suspicion of nature credits.

The direction of travel is right

04 · Data-driven methodologies as the higher bar

The wider market has moved the same way. The Integrity Council for the Voluntary Carbon Market assesses methodologies against its Core Carbon Principles, and VM0047 was approved as CCP-eligible, alongside the deforestation methodology VM0048, marking dynamic, remote-sensing-led approaches as the higher-integrity direction for nature-based crediting. The signal is unambiguous: the field is converging on measurement that is harder to game. None of what follows is an argument that VM0047 is weak. It is an argument about what even a strong methodology is structurally unable to do.

Three things a methodology cannot reach

05 · Independence

A dynamic benchmark is only as independent as the hand that builds it. The benchmark is constructed by, or on behalf of, the credited party, and the validation and verification body that signs it off is selected and paid through the same chain. Even sympathetic reviewers flag the seam. In its assessment of the methodology, BeZero Carbon welcomed the dynamic baseline as best practice and then noted that the matching uses a single factor, the stocking index, with no handling of uncertainty in the match, and that control-plot locations are not required to be disclosed for independent assessment. Its conclusion was explicit: independent risk assessment of individual projects and vintages remains key. A methodology can specify the maths. It cannot supply the independence of the party doing it.

06 · Permanence

VM0047 sharpens the count of tonnes added. It does not price the risk that those tonnes leave. Permanence in the VCS programme is handled separately, through the AFOLU Non-Permanence Risk Tool, which scores reversal risk and withholds a percentage of credits into a pooled buffer account. The buffer is a self-insurance pool, sized at a point in time from a risk rating, and Verra has itself acknowledged the gap by consulting on a long-term monitoring system to track reversals, including beyond the crediting period, and to adjust buffer withholdings against observed losses. The honest reading is that fire, drought and reversal are continuous, climate-correlated hazards, while the buffer is a periodic estimate. Counting carbon accurately at verification says little about whether the buffer is adequate against next season's fire.

07 · The book

The third gap is the largest. A methodology certifies a project at discrete verification points. Capital does not hold a project at a verification point. It holds a portfolio, continuously, across vintages, geographies and methodologies, exposed to reversal, downgrade and policy change every day in between. VM0047 produces an excellent measurement of one project at the moment of audit. It is silent on the correlated risk sitting across a book of assets between audits, which is exactly the read the holder of the risk most needs and is least able to get from the party being paid to issue.

Where the methodology ends and the read begins

VM0047 should be defended on its merits. It ended the developer-chosen baseline, it nets out the regeneration that the Saloum Delta showed was never the project's to sell, and it is rightly recognised as the higher-integrity path. But a better methodology improves the measurement. It does not change who produces it, who pays for it, or the fact that it certifies a project while capital carries a book. Those are not flaws to be patched in version 1.2. They sit one layer above the methodology, and that is where the risk actually lives.